Policing by Algorithm

The LAPD reduced the time needed to create 'Chronic Offender Bulletins' from one hour to 3-5 minutes using Palantir's Gotham platform ^[12][14].

The LAPD's Automated Bulletin Processing

Palantir's Gotham platform processes vast quantities of data from crime reports, automated license plate readers, and rap sheets to generate targets ^[11]. This rapid processing allows for near real-time generation of targets, creating an operational dependency on automated output ^[12][14]. The sheer volume and velocity of data processed can outpace the capacity of human agents to perform independent verification of surveillance leads ^[2][6].

The LASER and ELITE Scoring Systems

The LAPD's Gotham platform assigns "LASER scores" to individuals through a points-based system, where each police contact increases a person's point value ^[11]. Similarly, the ELITE system uses an "address confidence score" to identify individuals and "target rich" areas for deportation ^[2]. This automation of target identification, particularly when processing historically biased crime data, acts as an "operational architect" that constructs new investigative priorities and "automated suspicion," according to a 2025 paper in the American Sociological Review ^[1][7][19]. In New Orleans, a partnership with the New Orleans Police Department (NOPD) generated a list of "likely" offenders based on social ties and arrest records, while in Los Angeles, Palantir's software helped designate "chronic offenders," disproportionately targeting minority neighborhoods ^[9].

The Fourth Amendment and the Mosaic Theory

Legal scholars argue that the cumulative effect of data collection and analysis, especially in the digital age, invokes the mosaic theory and the Carpenter precedent ^[20][22]. Automating target identification through extensive personal data aggregation mirrors unconstitutional long-term tracking because of its capacity to reveal intimate details of private life without individualized suspicion ^[20][21][22]. This suggests that the use of automated scoring systems and aggregated data by law enforcement, particularly for identifying enforcement targets, is likely to be classified as a "search" under the Fourth Amendment, requiring a warrant ^[23][24][25].

The Expansion of the US Surveillance State

The software consolidates Social Security Administration (SSA), Internal Revenue Service (IRS), and Department of Homeland Security (DHS) records, creating a single dataset containing identity and financial data for effectively every American ^[9]. Palantir's integration expands surveillance to individuals without prior police contact ^[1]. This qualitative shift moves domestic policing from specific counterterrorism efforts toward daily law enforcement and immigration operations ^[3][7][12]. Palantir explicitly denies building any such "master database" and states that each customer instance of its software is legally, technically, and operationally distinct ^[10][11].

The Influence on Policing Decisions

Palantir's platforms accelerate target identification and resource allocation for agencies like the LAPD and Immigration and Customs Enforcement (ICE). Evidence suggests these systems create operational dependencies and can cause measurable shifts toward disproportionate targeting and reinforced racial bias, influencing individual policing decisions beyond mere acceleration. The speed of these systems can outpace human oversight, raising questions about the degree to which they dictate, rather than inform, real-time choices by officers, such as those concerning target identification for the LAPD.

Download the full research report (PDF)